Perhaps more than ever, high-end computing and telecommunications applications, for example, are using highly optimized integrated circuits such as microprocessors, field programmable gate arrays (“FPGAs”), application-specific integrated circuits (“ASICs”), etc., wherein the silicon processes for such circuits are selected and/or adjusted to maximize performance and reduce costs. This very often yields differing power supply requirements for each circuit, i.e., different discrete voltage and current levels. Further, many of these circuits require a relatively low voltage (e.g., 1 v or less), but with relatively high current (e.g., 100 A). It is undesirable to deliver relatively high current at low voltages over a relatively long distance through an electronic device for a number of reasons. First, the relatively long physical run of low voltage, high current lines consumes significant circuit board area and congests the routing of signal lines on the circuit board. Second, the impedance of the lines carrying the high current tends to dissipate a lot of power and complicate load regulation. Third, it is difficult to tailor the voltage/current characteristics to accommodate changes in load requirements.
Decentralized power architectures have been developed to address the power supply requirements for such systems. In one such power architecture, an intermediate bus voltage is distributed throughout the electronic system, and an individual point-of-load (“POL”) regulators, i.e., DC/DC converters, are located at the point of power consumption within the electronic system. Each POL regulator would convert the intermediate bus voltage to the level required by the corresponding electronic circuit. Ideally, the POL regulator would be physically located adjacent to the corresponding electronic circuit so as to minimize the length of the low voltage, high current lines through the electronic system. The intermediate bus voltage can be delivered to the multiple POL regulators using low current lines that minimize loss.
This decentralization process can be pushed so far that almost all loads (microprocessors, FPGAs, etc.) in the application have their own power supply. In other words, while an application (as a whole) may be designed to perform a single main function, its power supply system (which provides power to the chips located therein) may be built from individual, stand alone POL regulators. A drawback of such power supply systems, however, is that they fail to reflect the inter-dependency of the loads the POL regulators are supplying. For example, if one POL regulator fails, then the circuit supplied by this converter will also fail without notifying or disabling the other chips that are dependant upon the first circuit. This can result in unpredictable malfunction of the load or further damage to the power supply system by overstressing the other chips and/or their related POL regulators. Conventional power supply systems provide only very simple fault management in the form of power-good signals, which an application can use to determine a faulty power supply condition of a particular POL regulator. This is generally insufficient to provide system-level protection for the loads in case of a fault.
Thus, it would be advantageous to have a system and method for managing faults in a distributed power system having a plurality of POL regulators.